YETNAYET AYALNEH BOGALE
February, 2012
SUPERVISORS:
Dr. Ir. M. H. P. Zuidgeest Dr. O. Huisman
Structure: Case Study in Addis
Ababa, Ethiopia
Thesis submitted to the Faculty of Geo-Information Science and Earth Observation of the University of Twente in partial fulfilment of the requirements for the degree of Master of Science in Geo-information Science and Earth Observation.
Specialization: Geo-informatics
SUPERVISORS:
Dr. Ir. M. H. P. Zuidgeest Dr. O. Huisman
THESIS ASSESSMENT BOARD:
Prof.Dr.Ir. M.F.A.M. van Maarseveen Dr. Jing Bie ; University of Twente
Evaluating Transport Network Structure: Case Study in Addis Ababa, Ethiopia
Yetnayet Ayalneh Bogale
Enschede, The Netherlands, February, 2012
Disclaimer
This document describes work undertaken as part of a programme of study at the Faculty of Geo-Information Science and Earth Observation of the University of Twente. All views and opinions expressed therein remain the sole responsibility of the author, and do not necessarily represent those of the Faculty.
Addis Ababa, the administrative and financial capital of Ethiopia, is experiencing continued growth and change. Enabling change is more complex and has many ramifications that need to be optimized and balanced by promoting sustainable growth. To facilitate rapid change and development, the Ethiopian National Urban Transport Policy (ENUTP) requires ensuring adequate, efficient and high quality transport infrastructure to guarantee effective mobility of goods and people. Although the proposed ENUTP objective seeks to achieve these goals, there are challenges which slow it down, the most significant being inadequate infrastructure in the current network arising from certain constraints.
This research identified inadequate infrastructure by analysing the spatial mismatch within the current road network measured against Network and Transport indicators developed to evaluate how well the existing network serves prevailing demand for travel. Further, we have evaluated the existing network structure against pre-determined transport planning objectives that allow the realization of the ENUTP objectives including accessibility, equity and efficiency by using indicators such as road density, mobility and proximity indices.
The Spatial Mismatch Indicator, central in this research, is operationalized by comparing the traffic assigned on the ‘Reference Network’ that is developed by using the Euclidean distance between the centroids of the Traffic Analysis Zones (TAZ) with that on the existing network. We have identified, evaluated and interpreted the existing mismatch by developing Spatial Mismatch Indices (SMI) that were reckoned using three methods to assign the traffic demand to the available supply. These are Euclidean Network-Based Assignment, Real Network-Based Assignment and finally, Real Network-Based Assignment with Disaggregated TAZs. In all the three methods, the spatial pattern of travel demand on the network is predicted by adapting several components of the classical Four-Step Transport Modelling approach, widely applicable in analysis of large networks. Following the execution of the trip generation model using regression models, the trip distribution is done based on the Gravity Model. Later, we have assigned the trips on the network using a basic All-Or-Nothing (AON) assignment.
In the Euclidean Network-Based Assignment method, the traffic is assigned to the Euclidean Network corresponding to the directed straight lines connecting the OD pairs. In the Real Network-Based Assignment methods depicted above, the network based trip distribution matrix is assigned to routes connecting the O-D pairs using AON assignment. The spatial mismatch at TAZ level, then, is reckoned as the total number of trips passing through each TAZ using the reference and real networks. The results of the implementation allow us to identify and evaluate inadequacy of infrastructure in the evaluated network and later distinguish the missing road infrastructure that will improve the existing performance.
The research shows that there are inadequate levels of infrastructure in parts of the current road network of Addis Ababa, particularly the peripheral areas suffer from lack of roads and roads in the central areas have capacity limitation. The research has indicated that spatial mismatch and missing road infrastructure can be better identified and evaluated by using disaggregated and spatially equalized TAZs more efficiently. Further, recommendations are made to improve the current road network structure considering connectivity and accessibility objectives that enable the realization of the ENUTP. Depending on the cost of construction and the access they open for development, the identified missing infrastructure is compared with the existing network for meeting the said objectives and later prioritized to produce a sensible transport plan based on improving spatial mismatch for decision makers.
Key words: Network structure evaluation, Spatial mismatch, Addis Ababa.
I would like to take this opportunity to thank my supervisors Dr. Ir. M.H.P. Zuidgeest and Dr. O.
Huisman for their excellent guidance, critical comments and encouragement throughout my research time.
They have shared their knowledge unreservedly and this work is made possible as a result of their thoughtful steering through the course of the research. It is a wonderful experience working with them and I am grateful for their friendship and advice.
The data used in this thesis are gathered from cooperative offices in Ethiopia. I would like to give especial thanks to the Planning & ICT Department of ERA for welcoming me to their office and their kind help in readily availing available data. Also I would like to thank the staff of Afri Geo Information Engineering Plc, Ato Mohammed Nuru for his kind cooperation with data acquisition.
The unlimited cooperation of ITC Student Affairs Office, especially, Ms. Bettine Geerdink and Ms.
Theresa van den Boogaard have been marvellous during my stay. My dear friends who have been sharing their experience and thoughts with me throughout the study period have been wonderful. Specially, I would like to thank, my dear friend, Hari Krishna Dhonju, who has been very helpful whenever I experienced programming difficulties. I would like to thank everyone who has participated in making this thesis successful.
Most especially, I thank my family and my husband, Beteseb; words alone cannot express what I owe them for their continuous encouragement and patient love that enabled me to complete this thesis.
Beteseb, thank you for choosing to share your life with me.
Even though the help and participation of all was very helpful, but without Jesus none of it would have been real. Thank you Lord, no one has been patient with me, eased my pain and loved me as you have.
List of figures ... v
List of tables ... vi
1. Introduction ... 1
1.1. General Introduction ... 1
1.2. Motivation and Problem Statment ... 2
1.3. Research identification ... 3
1.3.1. Research Problem ...3
1.3.2. Research Objectives ...3
1.3.3. Research Questions ...3
1.4. Conceptual Framework ... 4
1.5. Thesis structure ... 4
1.6. Research Design ... 5
2. Literature Review ... 6
2.1. Introduction ... 6
2.2. Modelling Transport Network ... 7
2.3. Network Evaluation Indicators... 8
2.3.1. Mobility ...8
2.3.2. Equity ...9
2.3.3. Accessibility ... 10
2.3.4. Transport Infrastructure Availability ... 10
2.3.5. Spatial Mismatch ... 10
2.4. Transport Demand Modelling ... 11
2.4.1. Traffic Analysis Zones ... 12
2.4.2. Trip Generation ... 13
2.4.3. Trip Distribution ... 13
2.4.4. Traffic Assignment ... 14
2.5. Contribution to Economic Growth ... 14
3. Study Area... 16
3.1. Introduction ... 16
3.2. Population and Socio-economic Characteristics ... 17
3.2.1. Population ... 17
3.2.2. Income ... 18
3.3. Land Use ... 18
3.4. Transport and Road Network ... 20
4. Methodology and materials ... 24
4.1. Data Acquisition ... 24
4.1.1. Primary Data ... 24
4.1.2. Secondary Data ... 25
4.1.3. Data Preparation ... 26
4.3.1. Trip Generation ... 28
4.3.2. Trip Distribution ... 28
4.3.3. Traffic Assignment to the Links ... 31
4.4. Trips Through Each Zone ... 32
4.5. Disaggregated Zones ... 33
4.6. Summary of Methods ... 36
5. Evaluation of Existing Network ... 37
5.1. Transport Demand Modelling ... 37
5.1.1. Trip Generation ... 37
5.1.2. Trip Distribution ... 38
5.1.3. Traffic Assignment ... 40
5.2. Mobility Index ... 41
5.3. Road Density ... 43
5.4. Proximity to the Road Network ... 44
5.5. Spatial Mismatch ... 45
5.5.1. Euclidean Network Based Assignment... 45
5.5.2. Real Network Based Assignment ... 47
5.5.3. Real Network Based Assignment with Disaggregated Zones ... 49
6. Quantification of Inadequacy within the Network ... 50
6.1. Identification of Inadequate Transport Infrastructure ... 50
6.2. Identification of Missing Infrastructure ... 52
6.3. Comparison of Current Network Structure with the Recommended Network Using Network Structure Indicators ... 54
7. Discussion and Conclusions ... 56
7.1. Discussion ... 56
7.2. Limitations of the study ... 58
7.3. Conclusion ... 58
7.4. Recommendations ... 60
List of references ... 61
Appendix A: Codes Used ... 64
Appendix B: Socio-economic data ... 67
Appendix C: Interview Questions ... 72
Appendix D: Maps ... 73
Appendix E: Input Structure ... 78
Figure 1-1) Conceptual Framework ... 4
Figure 1-2) Research design ... 5
Figure 2-1) The classic four stage transport modelling(Ortúzar & Willumsen, 2011) ... 12
Figure 3-1) Administrative units of Addis Ababa ... 16
Figure 3-2) Income levels and population density ... 18
Figure 3-3) Land use ... 19
Figure 3-4) Road network and hierarchy ... 22
Figure 3-5) Traffic Analysis Zones ... 23
Figure 4-1) Attribute of desire lines ... 27
Figure 4-2) Real and Reference road networks ... 27
Figure 4-3) Attributes of reference network. ... 31
Figure 4-4) Population distribution among disaggregated zones ... 35
Figure 4-5) Summary of Methods ... 36
Figure 5-1) Total trip generation ... 38
Figure 5-2) Desire lines and Real network ... 39
Figure 5-3) Links with trip greater than 1000 trips/ hour ... 40
Figure 5-4) Trip assignment on the real network using all modes (trips per hour) ... 41
Figure 5-5) Links with Mobility Index greater than 1.41 ... 42
Figure 5-6) Road density, Road length per population in the left (km/1000person) and Road length per area in the right (km/km2) ... 43
Figure 5-7) Income level and proximity to roads ... 44
Figure 5-8) Proximity to road infrastructures ... 45
Figure 5-9) Trips passing through each zone by assigning trips on the reference network ... 46
Figure 5-10) Spatial Mismatch Indices using original method ... 46
Figure 5-11) Trips passing through each zone by assigning Euclidian based trips on the reference network (left) network based trips on real network (right). ... 47
Figure 5-12) Spatial mismatch indices using the improved method ... 48
Figure 5-13) Spatial mismatch indices for disaggregated zones ... 49
Figure 6-1) SMI 1 threshold values ... 51
Figure 6-2) SMI 2 threshold values ... 51
Figure 6-3) SMIs and Corridors Identified for Improvement ... 52
Figure 6-4) Identified Missing Links ... 53
Figure 6-5) Recommended Road Network ... 54
Table 3-1) Population of Addis Ababa as compared to Ethiopia ... 17
Table 3-2) Addis Ababa Population by Sub-City (source inception report) ... 17
Table 3-3) Land use proportions (source UTS (2004/2005)) ... 19
Table 3-4 ) Road hierarchy (Source AACRA) ... 21
Table 3-5) Transport modes and percentages (Source UTS (2004/2005)) ... 21
Table 4-1) Data required ... 24
Table 4-2) Secondary data collected and sources ... 26
Table 4-3) Regression equation for Trip attraction and production( source Urban Transport Study (2004/2006)) ... 28
Table 4-4) Gravity model calibrated parameters(source Urban Transport Study (2004/2006)) ... 29
Table 4-5) Occupancy and equivalency factors (Mekbib, 2007) ... 32
Table 4-6) Land use ranks ... 34
Table 5-1) Trip Share by purpose ... 37
Table 5-2) Trip shares of Sub-cities ... 38
Table 5-3) Road Types and Capacity ... 41
Table 5-4) Mobility Index ... 42
Table 6-1) Network structure indices ... 55
1. INTRODUCTION
1.1. General Introduction
With the ever increasing level of urbanization, the issue of developing an efficient strategy for urban transportation has been considered in numerous scientific and technical works, especially in the context of developed countries (Pratelli & Brebbia, 2010). The definition and eventual appraisal of various strategies for the future development of urban areas represent a demanding task that requires versatile experience in traffic engineering, planning and economics. Eventually, the decision-maker needs to be aware of the multi-dimensional impacts of various planning and management measures on the efficiency of the transport system towards meeting certain objectives to make an informed choice between the available options (Keshkamat, Looijen, & Zuidgeest, 2009).
When we look at the transport planning and decision making strategies of cities of developing countries most of the times, planning decisions are made based on pure speculation and it is hard to explain how decisions are made that affect the road network plan. Due to this fact cities grow in uncontrolled manner and evolve into more and more inefficient transport networks. It is clear that these cities need a strategic plan that deals with the overall structural and capacity of the road network and with the transport land use interaction in urban areas.
The essence of urban transport planning is to provide adequate and equitable recourse among the population. In developing countries with limited funds to distribute resources equally and to stick within the available funds planning strategies that identify most in need areas and prioritising these areas is needed. One of the methods that allow us to identify the unprivileged and neglected area is by evaluating the current transport supply with respect to demand.
Many researchers agree that major changes in transport network structure affect patterns of urban development and location of social and economic activities, households and employment centers. In the other hand major change in land use influence the trip making behavior of people, destination and mode chose (Waddell, 2011). The network structure also affect accessibility and destination choice of travelers (Huang & Levinson, 2011).For developing cities where the road network is the dominate mode of transport land use transport relationship is highly coupled. The road network provides access to different activities like businesses, education and employment opportunities (Murray, et al. 1998) and people make trips to join these activities. It has major influence in the economic development by opening virgin lands for agriculture and deliver agricultural products to the market, it serves as freight transportation for land locked countries.
However the road network in most developing countries suffers from many problems like; high accident levels, inadequate infrastructures both in capacity and availability, poor quality infrastructure and mismatch between demand and supply. These are caused by high urbanization, city growth and lack of proper transport planning strategies. Even though no transport network can serve all travel demand perfectly, the amount by which it fails to do so can be useful to study existing road network and identify areas with inadequate infrastructure (Davidson & Davidson, 1998). In this research context inadequacy can be distinguished in two ways:
1) Infrastructure exists but has low capacity and
2) Infrastructure does not exist at all
The current performance can be measured by assessing the accessibility, level of mobility provided by the network and by analysing the efficiency of the network in comparison with ideal network.
Accessibility in this context is ease to reach certain destinations from a particular origin using transport network and efficiency in terms of minimized cost. Geographic Information Systems can be used for better understanding of accessibility and also be used in automatic identification of missing infrastructure in existing network structures (Zuidgeest, Rouwette, & Jong, 2009).
1.2. Motivation and Problem Statment
In light of the fore-going, Addis Ababa, the administrative and financial capital of Ethiopia, is experiencing continuous growth and change. Change is experienced in all dimensions of the city but different parts of the city grow at different rates. Economically, the city is transforming from a predominantly administrative and service center into an industrial and financial center. Due to the rapid economic growth and change, there is high mobility of goods & passenger which leads to high transportation demand. However, the existing transport system is characterized by high accident levels, high traffic congestion, negative environmental impacts, unsafe public transport and low accessibility levels. Main causes for these problems link with poor infrastructure including road and public transport network, poor interaction between land-use and transport planning, inadequate road infrastructure and low transport network density.
The transport sector being the backbone to the economic growth of the nation, road network remains the basic and critical component of transport system in the city. Addis Ababa has a radial form of road network which is shaped by five major roads radiating out of the Central Business District (CBD) into the outskirts. However, the recently constructed Ring Road has added an orbital road around the periphery of the CBD. There are other road links which in their continuity can be considered as partial orbital corridors. Even though a well-defined hierarchical system is missing, still the road network could be classified into a hierarchical system comprising of arterial, sub-arterial, collector and residential access roads. As described in the Urban Transport Study (2004/2006) the road network in Addis Ababa suffers from many inadequacies including absence of a balanced hierarchical system, capacity limitation, absence of infrastructure and low density among others.
There are not many studies undertaken to analyze the transportation problem for Addis Ababa transport network and where the network design itself may be the cause of the problem. However, some published sources which analyze the transportation problem of the city include Urban Mobility in Three Major Cities by the World Bank (The World Bank, 2002 ). In this study, the transport system of the city was analyzed and remedial measures are proposed to improve the performance of the transport system and the future planning and management of transport infrastructure. Once again, The World Bank, in collaboration with Addis Ababa Roads Authority and Ethiopia Road Authority (Ethiopian Roads Authority, 2004/2006), has clearly noted that the current road network is inefficient and measures should be taken for improvement in capacity, efficiency, etc. However, both these papers did not analyze the problem based on an objective evaluation of the spatial structure of the existing transport network.
The vision of transport plan for the city is “affordable transport, enhanced access and mobility”. The world bank report on Three Major Cities (The World Bank, 2002 ) has noted that even though the city has low urbanization in the continent, it suffers from urban mobility problems. Due to population growth the city expanding rapidly but the road network fails to serve the newly developed areas. The
public transport is limited to certain parts of the city due to unavailability of road network. So most of trips are made are by foot and restricted to some regions. In the most of peripheral areas of the city due to absence of road infrastructure accessibility is very low. Some of these areas are new developed real estates and areas that provide agricultural products to the city. As pre the UTS (2004/2006) plan some of these areas are reserved for future industrial development.
1.3. Research identification 1.3.1. Research Problem
The above sections reveal that the current transport infrastructure of Addis Ababa may be inadequate with respect to travel demand. GI based network indicators can be developed to assess and evaluate inadequacy within the network structure. From the literature available, there are established indicators that can assess networks in general, but most of them do not consider travel demand. Further, limited GI based network indicators are available to identify inadequate infrastructure in an existing road network. GI based network indicators can be developed to assess network structure. Hence these indicators can identify infrastructure with low capacity and missing connections. As such, improvement should be recommended in line with the vision of the city ‘affordable transport, enhanced access and mobility’.
1.3.2. Research Objectives
To develop GI based indicators to evaluate road network structure in relation to potential travel demand and recommend improvement in network structure.
Sub-objectives:
1. To identify potential travel demand and current network.
2. To develop indicators to assess the existing road/transportation network structure.
3. To identify which network based indicators can be used to identify inadequate infrastructure for prioritization and define road hierarchy for Addis Ababa.
1.3.3. Research Questions
The research objectives identified above are translated into the following specific research questions:
1. To identify potential travel demand in current network.
What models can be used for estimation of travel demand in urban area?
2. To develop indicators to assess the existing road/transportation network structure.
Which network based indicators should be developed to assess the structure of current transport network?
How to evaluate the current network based on the pattern of travel demand?
How to combine network indicators and spatial nature of demand to evaluate structure of transport network?
3. To identify which network based indicators can be used to identify inadequate infrastructure for prioritization and define road hierarchy for Addis Ababa.
How to identify inadequate infrastructure in existing road network through assessment of the network indicators?
How to identify inadequate infrastructure based on pattern of travel demand and available transport network supply?
How to prioritize infrastructure based on their rank in achieving pre-determined transport objectives?
1.4. Conceptual Framework
Transport system is combination of demand and supply. Travel demand is generated by the desire of people to join activities. It depends on the socio-economic activity of the people, land use characteristics and accessibility of zones. Trips made also vary depending on the purpose of the trip, time at which the trip is made, person type; socio-economic activity like car ownership, income and household size, and the modal choice. From Urban Transport Study (2004/2006) trip by purpose (employment, home based, education and other trips) is considered. Transport supply includes infrastructure, services and transport networks. For this research we will only consider the transport network. Combining travel and transport supply, the network structure will be evaluated that requires us to set up criteria that can be used for this purpose.
Road Network
Land use Zone map
Socio-economic and Population data
Network Evaluation Select criteria and Network based indicators
for each criteria
Evaluate the network structure
identify inadequate infrastructures
Supply Demand
Figure 1-1) Conceptual Framework 1.5. Thesis structure
The overall thesis is organised under six chapters as follows: Introduction, Literature review, Study area description, Methods, Results and Discussion and Conclusion and Recommendations.
Chapter 1gives brief introduction of the research, identifies research problems, defines research objectives and questions related to the objectives.
Chapter 2 briefly review literature on transport network modelling, criterion that are used to evaluate transport network structure and travel demand modelling.
Chapter 3 describes the study area based on its topography, socio-economic characteristics, demography and land use
Chapter 4 presents the research approach and data collection methods used to reach the objectives of this research are explained.
Chapter 5 discuss the results of transport modelling and evaluation of the current network. Furthermore results of the spatial mismatch assessment are provided.
Chapter 6 identification of in adequate infrastructures are discussed and interpreted.
Chapter 7 provides the discussion, conclusions and recommendations made based on the results of the research.
1.6. Research Design
The following flow chart depicts the logical steps carried out in the course of the research.
Inadequate infrastructures
Network Evaluation Data Organization
Problem Identification
L i t e r a t u r e
r e v i e w
Identification of data required
Primary Data
x Interview
Secondary Data
x Land Use map
x Traffic Analysis Zones
x Road network
x Socio-economic data Formulation of research objectives
and research questions Problem Definition Real world
NBI for low capacity infrastructures
Identify low capacity infrastructures
NBI for missing infrastructures
Identify missing infrastructures
Format Conversion
Editing and adding missing Attributes
Joining tables with
maps
Recommend Road Network
Compare Current and Recommended
Network
Conclusion and Recommendations
Network Analysis
Evaluating the network structure
Transport modelling Set criteria
Network based indicators
Figure 1-2) Research design
2. LITERATURE REVIEW
2.1. Introduction
One of the factors that affect urban form to a great extent is transport network structure. The settlement of people may be depending on the availability of transport infrastructure or transport infrastructure may be constructed where demand is at most. In either way they affect each other. The availability and quality of road infrastructure affects the quality of life, economic and business activities and strengthen the economy of a region. Transport infrastructure plays important role in the economy; especially for developing countries it opens access to agricultural lands, markets, health centers, schools and so on. All in all, it facilitates mobility in urban areas and improve quality of life (Gwilliam & Kenneth, 2002).
There are many important decision-making problems in transport network planning. One of them is due to the nature of transport networks projects that are planned for long-term, decisions on investment also need long-term perspective (Santosa & Joewono, 2005). Especially developing countries with limited funds need to spend the money effectively and satisfy the areas which are in need. Evaluating the current road network with respect to the current demand will help decision makers to understand how much the network satisfies the demand and identify which areas need more attention for future development plan.
The current performance of road network, can be evaluated based on the structure of the road network (Xie & Levinson, 2007), the access and mobility it provides (Gutierrez, el al.,1998; Liu & Zhu, 2004) demand and supply equilibrium (Bell & Lida, 1997) by measuring the capacity of the network (Chen et al., 1999) and by assessing the spatial mismatch between demand and supply(Grishchenko, 2011).
Even though no transport network can serve all travel demand perfectly, the amount by which it fails to do so can be useful to study existing network and identify areas with inadequate infrastructures (Davidson & Davidson 1998). The absence of one or more network links, especially those that have high travel demand, could lead to overall poor performance of the network in terms of overall system travel time and optimization of supply (Bell, 2000; Chen et al. 2002; Smith et al. 2003 cited by (Scotte et al.
2006)). Network effectiveness indicators, which compare the real road network to geographically perfect network (direct link), can be used to identify these areas (Davidson & Davidson, 1998).
In the interest of improving the performance characteristics of a network, a combination of demand and network structure related indicators are required. From the previously developed methods, even though there are GI based indicators that are useful to study transport network structure, the ones that consider travel demand are few. Those that consider spatial behavior of demand have been applied to bicycle networks (Zuidgeest, el al.,2009), freight networks (Grishchenko, 2011) and public transport (Abate Abreha, 2007). Further, limited methods are available to identify and prioritize inadequate infrastructure in existing road network.
In this study, the current transportation network structure of Addis Ababa will be evaluated by using network indicators to analyze the existing pattern of travel demand and the structure of the network supply. The literature review is organized in such a way that; it first briefly reviews transport network modelling, second part identifies indicators to evaluate the transport network, in the third part Travel
Demand Modelling will be discussed and at last, transport network contribution to economic growth will be discussed.
2.2. Modelling Transport Network
Transport demand modelling will help us in understanding the travel pattern, which itself provides an understanding of the relationship between urban structure and the transport network (Timmermans et al., 2003). Travel demand takes place at a certain time and space. To capture the spatial and temporal nature of travel demand the supply side of transport system (road infrastructure) can be represented by network.In the past decades, advanced methods have been developed in order to analyze more complex transportation network problems and propose solutions. These models can be used in either of the following types:
1. Static (relating to states or conditions) or dynamic (relating to processes), 2. Computational or non-computational,
3. Quantitative (with numbers) or qualitative (without), 4. Spatial or non-spatial,
5. Empirical (based directly on data) or synthetic (based on a relationship derived from the data) (Ortúzar & Willumsen, 2011).
The use of each type depends on, among others, the context of analysis, the purpose of the exercise, accuracy and certainty required, the level of detail, statement of the problem, availability of computing tools and availability of time and resource to carry out the analysis (Button & Hensher, 2000). Ideally, we would like the model to be appropriately detailed, accurate, appropriately sensitive, economical and easy to use. However, very few models have all these qualities; trade-offs have to be made in the light of the requirements of the exercise (O'Flaherty, 1997); (Bruton, 1993). To analyze real network considering land use and socioeconomic data, a model which can handle large spatial data is needed. Even though transport is dynamic in nature modelling large real networks usually is done on the basis of static models, as is done in this study.
From these models, GIS is able to handle huge data and analyze a variety of network-related problems by itself and can also be integrated with other methods for transport network analysis by providing a spatial database and mapping platform (Kuby, et al., 2005). For instance, it can be integrated with Graph Theory to measure network efficiency (Rodrigue, 2009).
Graph Theory depends on the concept of representing networks as a graph or matrix. The underlying basics of this science assumes that transport networks can be represented by directed graph with nodes and links where the nodes represent junctions while links indicate homogeneous road sections between nodes (Zuidgeest & Maarseveen, 2011).These network analysis methods are founded on the principle that the efficiency of a network depends partially on the geographical lay-out or structure of the nodes and links forming the network. The matrix representing the network can be manipulated mathematically with a series of network measures (Kuby, et al., 2005).
Since the late 1950s, several network based indicators/measures have been developed to analyze transport network based on structural efficiency, connectivity, cyclic property, etc. of the network. For example, Garrison and Marble ((1962), (1964) and (1965)), developed the first of such indicators, including:
Alpha index (a measure of connectivity which evaluates the number of cycles in a graph in comparison with the maximum number of cycles),
Beta index (which measures the level of connectivity in a graph and is expressed by the relationship between the number of links over the number of nodes) and
Gamma index (measure of connectivity that considers the relationship between the number of observed links and the number of possible links).
Recently, Xie & Levinson (2007) developed new indicators which consider flow on the road network for measuring the structure of road network. Even though these indicators consider flow, they do not consider spatial distribution of demand. The other difficulty is, indicators can differ significantly depending on many factors such as city type, regional transportation vision, or travel behavior even between modes (Derrible & Kennedy, 2011). Finding appropriate indicators that can be used to assess transport network structure of Addis Ababa is one of the interests of this study.
In urban transport, the network structure and flow mutually affect each other (Xie & Levinson, 2007).
Therefore, while optimizing the network, a combination of spatial (with respect to demand) and network indicators is required. Network indicators analyze the network structure whereas the travel demand (trip) consists of the desire to make a trip. Trips are made to join activities via particular modes of travel, which uses specific routes through the transport network. Once this demand is realized, it becomes spatial interaction which flows through transport network.
To capture the spatial nature of demand, sciences like Transport Geography which considers the spatial perspective of socio-economic, industrial and settlement frameworks (which are main factors of transport demand) within the transport network development and transport system operates (Hoyle &
Knowles, 1992) can be useful. This field analyzes land use-transport interaction as spatial interactions in transport are derived from land use (Hensher, el al.2004). The transport network considers spatial organization of the transport infrastructure and terminals as its basic structural elements, links and nodes respectively (Taaffe, Gauthier, & O'Kelly, 1996).
In this study, the current transportation network structure of Addis Ababa will be evaluated by using network indicators to analyze the existing pattern of travel demand and the structure of the network supply. A methodology which will assist in rationalizing the network by identifying and prioritizing the required infrastructure is developed and implemented in for Addis Ababa Transport Network. The recommended transport network that follows will be compared with the existing one based on pre- determined transport planning objectives.
2.3. Network Evaluation Indicators
Evaluation of transport system should be based on defined development objectives of the city.
Evaluating criterion is set depending on the goal of development of each city. Criteria and indicators may differ significantly depending on many factors such as city type, regional transportation vision, or travel behaviour even between modes (Derrible & Kennedy, 2011).
2.3.1. Mobility
Transport networks are intended to move people and goods to where they need to go quickly and affordably(Sohail, Maunder, & Cavill, 2006).The movement of people and goods is affected by cost and
the safety of travel. Mobility is the ability to move people and goods. Increasing the efficiency and effectiveness of transport network will increase mobility. In developing countries with low infrastructure both cost of travel and safety are major factors affecting mobility. Mobility measures indicate the quality of movement and the quantity being moved (Zuidgeest, 2005).
Efficiency of a network can be indicator for the mobility offered by the network. An ideal road network is the one, which provide the most direct route between an origin and destination at the desired speed.
In order to measure the efficiency of the road network, a mobility index is defined. Mobility index is defined as the ratio of travel time by the physical route (speed determined by the type of the road) between an origin and destination and the travel time by the airline distance at desired speed (Ethiopian Roads Authority, 2000/2001).
𝑀𝑜𝑏𝑖𝑙𝑖𝑡𝑦 𝐼𝑛𝑑𝑒𝑥 (𝑀𝐼) = 𝑇𝑟𝑎𝑣𝑒𝑙 𝑡𝑖𝑚𝑒 𝑏𝑦 𝑡ℎ𝑒 𝑝ℎ𝑦𝑠𝑖𝑐𝑎𝑙 𝑟𝑜𝑢𝑡𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑜𝑟𝑖𝑔𝑖𝑛 𝑎𝑛𝑑 𝑑𝑒𝑠𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛
𝑇𝑟𝑎𝑣𝑒𝑙 𝑡𝑖𝑚𝑒 𝑏𝑦 𝑡ℎ𝑒 𝑎𝑖𝑟𝑙𝑖𝑛𝑒 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 (2.1
Mobility is affected by the network structure and condition. Network structure determines how direct the route between two points is and type of the road determines how fast the vehicles can travel. The network will be considered well connected if the mobility index falls between 1 and 1.41 (the ratio of the sum of the two sides of triangle and the diagonal)(Ethiopian Roads Authority, 2000/2001). If the road network can provide a mobility index in this range for travel between all regions, then the road network is considered to have provided excellent level of connectivity and mobility.
2.3.2. Equity
Equity refers to the distribution of resources (in this case transport network) and if the distribution is considered appropriate (Litman, 2011). In transport equity is a diversified concept and the analysis may be difficult because there are various ways to categorize people, number of resources to consider and various ways measuring these resources. Broadly speaking equity in transport can be categorized in to two types (Geurs & Ritsema van Eck, 2001).
1) Horizontal Equity: focuses on the distribution of resource between individuals or groups with comparable needs and abilities. Equal individuals or groups should share equal resource, pay equal cost and must be treated in the same way.
2) Vertical Equity: is concerned with the distribution of resources between individuals or groups that differ in ability and need. From this definition transport networks are equitable if they favour socially and economically disadvantaged groups. Social, Economic and Spatial equities are considered.
The development plan of a city should be prepared with due consideration of equity requirements. In developing countries with different nationalities like Ethiopia, equity between different ethnic groups must be considered. Based on the development plan for Addis Ababa, in this research we will consider Spatial and socio-economical equities.
In planning road networks, it is very important that the road network is equitably developed in all zones, in order to promote general economic and social development objectives. The main consideration in the evaluation of new road projects is generally the economic viability. As a result, less developed areas with low traffic generation are often either not taken up or given low priority, further increasing the disparity
in development. There should be equitable development of the road network to ensure that the road network will facilitate the development of all zones.
2.3.3. Accessibility
Accessibility is one of the important characteristics of urban transport and it shows the relationship between transport and land use(Liu & Zhu, 2004). In the past years researchers have used accessibility for integrated transport land use planning for urban areas(Liu & Zhu, 2004) ,as a key element in efficiency analysis of transport network and infrastructure planning (Gutierrez, et al., 1998), and to generate the travel demand on public transport (O'Sullivan, et al. 2000).Curl, et al (2011) have revised literatures on the theoretical definition and measures of accessibility and in what extent it can be used to reduce inequality in the society. It can also be used to access changes by providing new road infrastructures(Liu & Zhu, 2004). Linneker & Spence (1996) used accessibility to analyse the impact of motor way development in regional development.
One of the goals of transport system is to increase mobility and access to facilities. Even though the definition of accessibility varies depending on its application(Curl, et al., 2011), based on this context, it can be generally defined as the ease with which certain destinations can be reached from a particular origin using specific mode of transport (O'Sullivan, et al., 2000). It can be measured by the mobility the transport system provides and density of opportunities that can be accessed within a certain time or distance. It depends on the spatial distribution of activities, the origins of demand and transport system connecting the origins and destinations.
The other important function of the road network is to provide linkage and open access to major centers, by giving connectivity to all economic centers, to sub city headquarters and national and international entry points like airports and freight hubs. By providing good connectivity the road network will contribute to the economic growth, increase accessibility and people mobility.
2.3.4. Transport Infrastructure Availability
One of the important factors affecting the performance of the transport network is the availability of adequate transport infrastructure. Transport infrastructures include road length, road width, public transport hubs and even road furniture and these factors affect the accessibility of transport system.
Especially road length per area and road length per unit population indicate the accessibility of the road network. This indicator is rather important in the context of Addis Ababa because, as is the case with many developing cities, given the deprivation of the society from enjoying acceptable levels of transport accessibility, some areas may be justified to have a specific link constructed purely on social basis as opposed to economic ones.
2.3.5. Spatial Mismatch
The idea of spatial mismatch was first developed in U.S.A in the late 1960s’ to assess the availability of jobs in black neighbourhood. Since then it has been used by different authors to detect the availability of jobs for the low income group within their reach depending on the cost of travel (Joseph, 2011; S McLafferty, 2001)and to analyse transport mode chose and ethnic groups(Patacchini & Zenou, 2005).
Considering the jobs as supply and employees as demand, this concept is used to analyse the spatial difference that exists between demand and supply.
The concept of spatial mismatch has not that much been applied in evaluating transport network evaluation. Grishchenko(2011) used this indicator for freight network assessment in Europe. This indicator compares trip distribution among traffic analysis zones considering the resistance (impedance) travellers experience by using the existing real network and that experienced in the direct route (Euclidean distance between two zones).
The trips along TAZs are trips that use the links in these zones. This will be computed using both resistances experienced using directed route (Euclidian network) and real network. From Rouwette (2005) cited by Grishchenko (2011)the spatial mismatch function is defined as the difference (∆Tij1) between network based (∆TijND) and Euclidean based (∆TijED) trips equation (2.2) and the second function takes the ratio between them (∆Tij2) equation (2.3).
∆Tij1 = ∆TijED - ∆TijND (2.2) Spatial mismatch index 2
∆Tij2 = ∆TijED/ ∆TijND (2.3) Spatial mismatch index 1 2.4. Transport Demand Modelling
Transport demand modelling was first developed in U.S.A during the 1950’s. Its’ important techniques were developed in mid 1970s. After years of research and experimentation it is recognized as part of transport planning process (Ortúzar & Willumsen, 2003). Since then it has been developing to solve many transport problems by providing information for evaluation, development and implementations of future transport planning proposals.
Travel demand modelling is one of the important part of the decision making process in transport (Button & Hensher, 2000). It assists engineers and planners to improve road networks, better utilize current network capacity , understand special impacts (Mark Zuidgeest & Maarseveen, 2011) .
Demand is driven from the need for travel of people. People will travel to reach to different activities from the place of residence to their jobs, shopping areas, schools or even others. Travel demand models should reflect the reasons for travel is to make part in activities and specially activities that are not present at the current position (Button & Hensher, 2000). The spatial separation of activities over space makes demand to take place over space. To deal with the spatial nature of demand is to divide study areas in to zones together with transport network (Ortúzar & Willumsen, 2003).
There are number of travel demand models used in urban transport studies. In this research we will adopt some parts of the Four Stage Urban Travel Demand Model. This model is the conventional method for Urban Transport Planning System (UTPS), where the distribution of land use in terms of population and employment allocation is done exogenously. This modelling approach is popularly known as sequential travel demand modelling which has four stages (refer figure 2.1), namely;
1. Trip Generation 2. Trip Destribution 3. Modal Split and 4. Trip Assignment
The approach considers zoning and network system, and the collection and coding of planning calibration of validation data (Ortúzar & Willumsen, 2011). The base year data include population of
each zone and level of economic activities. It establishes quantifiable relationships between travel pattern and population, spatial distribution of economic oppurtunity (employment) and socio economic characteristics of the population in the study area.
Figure 2-1) The classic four stage transport modelling(Ortúzar & Willumsen, 2011) In this research, we will use the several components of the Four-Step Transport Model (FSTM) for the analysis of travel demand on the road network. As standard traffic modelling exercise on large networks suggests, the study area will be partitioned into TAZ to quantify the demand using the first sub-model of the FSTM, namely, Trip Generation model will be used to estimate the correlation between socio- economic properties of the TAZ and levels of trip productions and attractions from them. The second sub-model of the FSTM, namely, Trip Distribution is, then, implemented by estimating the zone-to-zone cost matrix for the network by using trip distance as the only factor contributing to the generalized cost of travelling between the TAZ centroids. The resulting Origin-Destination matrix forms the trips that are made between the TAZ centroids. Since, at this stage, we will consider all modes of transport, to make the trip, therefore, the modes will not be split and the OD Matrix will have considered all the available modes. At this stage, the third sub-model of FSTM, Modal Split shall have been modelled and implemented to estimate the trips that are made using the different modes of travel. We have adopted the modal split results obtained from the Urban Transport Study (2005) for the purpose instead of modelling our own Modal Split Analysis due to lack of data and limited time for the research. The last of the Sub-Models of the FSTM is Traffic Assignment stage where we assign the trips by modes to the available network supply. The research has used the All-Or-Nothing (AON) Traffic Assignment for this purpose. The AON Assignment is the simplest of available assignment models which, however, offers the best estimate of assigned traffic on the network in a case where the trips are heavily dominated by walking modes, as is the case in the study area under consideration. Finaly trips passing through each zone will be computed by corrosponding OD routes passing through them. The following section discusses these transport modelling steps in greater detail.
2.4.1. Traffic Analysis Zones
Urban transport modelling depends on traffic analysis zones (TAZs) as its basic unit of analysis(You, Nedović-Budić, & Kim, 1998).The centroids of TAZs are used to represent trip origins and destinations.
Every employment centers, shopping centers, households and other activities of the planning region are aggregated into zones and are further simplified in to single node assuming they are concentrated at the centroid.
Before performing a transport model, we must decide the level of detail to be adopted in the study depending of the accuracy needed and resources available. A greater accuracy can be achieved by using more detail zoning system because this would eventually represent every individual. However, using highly detailed zones may not be economically feasible since handling large volume of data may be difficult whenever forecasting is involved (Ortúzar & Willumsen, 2003). The choice of TAZ size also depends on the type of analysis to be undertaken and the statement of problem involved. For instance, studies for traffic management and corridor choice need fairly disaggregated zones, requiring the introduction of many zones for the analysis whereas larger zones can be used in strategic studies that occasionally do requires lesser number of zones to be dealt with under the study.
2.4.2. Trip Generation
Trip generation and distribution are in the first step of transport modelling where the characteristics of the traveller and land use activities are evaluated, calibrated and validated to produce non-equilibrium measure of travel demand (Button & Hensher, 2000).The main purpose of this step is to process and estimate the total number of trips generated and attracted by each area unit (zone) in relation with the land use and the socio-economic characteristics of each zone(Ortúzar & Willumsen, 2011). Trips made may vary depending on the purpose, time of the day and person type. Trips by purpose include; work trips, education trips and others. Based on the time of the day trip is made it can be categorized as peak and off-peak. Socio-economic character is also another factor that affects the trip making behaviour of a person.
This model estimate the total number of trips produced or originated from a zone and attracted to each zone. There are three approaches commonly used in the trip generation analysis: regression analysis, trip rate analysis, and cross-classification analysis. These approaches establish statistical relationship between number of trips produced and land-use characteristics of the zone and socio-economic character of households.
2.4.3. Trip Distribution
The next step in travel demand modelling is trip distribution. Since we know total trip production and attraction potential of each zone, in this stage, the next information we want to know is where this trips go to and where the attractions comes from. Trips are distributed over the zones depending on the cost (impedance). Impedance may be time, money or distance or even combination of factors (Mark Zuidgeest & Maarseveen, 2011). This model is destination choice model that generates trip matrix (O-D matrix) Tij for each trip purpose utilized in the trip generation model and network attributes (inter-zonal impedance)(Button & Hensher, 2000). There are many trip distribution models, however, for this research Gravity Model is adopted.
Gravity Model
This model is mostly used when base year OD is not provided or important changes take place in the land use and transport network. The model is mainly based on Newton’s Gravity Low and assumes trip making behaviour is influenced by external factors like total trip ends and distance travelled (Ortúzar &
Willumsen, 2003). After many experiments and researches it was concluded that the effect of distance on trip making could be modelled better by decreasing function with the following equation.
𝑇 = α𝑂 𝐷 𝑓(𝑐 ) (2.4)
Where: 𝑂 and 𝐷 are the number of total trip ends in zone i and j respectively.
α is balancing factor
𝑓(𝑐 ) is the deterrence function or impedance function. This function decreases as travel cost or distance (time) increases. There are three versions of impedance functions that are usually used. In the UTS(2004/2006) the deterrence function used is combined function.
𝑓(𝑐 ) = (𝑐 )𝑒 ∝ Combined deterrence function (2.5) Where: 𝑐 is the generalized cost, time or distance of travel between origin i and destination j.
2.4.4. Traffic Assignment
This step is performed in the subsequent part of transport modelling exercise where demand is loaded to the transport network. There are number of traffic assignment models developed in the past years. The simplest route choice model is all-or-nothing assignment. In this assignment, all the trips from any origin to any destination is assigned to single minimum cost path between them (Ortúzar & Willumsen, 2011).
This method assumes that all trip makers are aware of the shortest route before making the trip and cost of travel stays the same. Traffic is assigned to links without considering the capacity of the link and congestion levels. This method has some limitations because it ignores the fact that cost on link is a function of volume and that when there is congestion multiple paths may be used.
Mekbib (2007) has noted that, simplified models like all-or-nothing (AON) assignment can be useful for developing countries like Ethiopia where versatile packages for traffic modelling and spatial data analysis are lacking for long term planning. AON assignment is also useful in areas where major mode of transportation is walking. In this study all-or-nothing assignment will be carried out by developing computer program. The step adopted is as follows.
1. Input data consist of network topology, link length, OD matrix converted to array.
2. For each origin-destination pair, find the shortest route
3. Assign the volume to the links forming the shortest route between the OD pairs 4. The output: Link number and volumes on each link
2.5. Contribution to Economic Growth
Transport is a key requirement for economic and social development to take place. Its’ absence causes isolation, backwardness and poverty. The World Bank World Development Report (2002)on ‘Urban transport and poverty reduction’ notes, that in some cases halving of transport costs has increased volume of trade by a factor of five. Equally dramatic examples could be taken from improving social welfare and skill levels through improved health and education.
“Transport infrastructures are, if not the engine, then the wheels, of economic activity” (The World Bank, 1994).It raises the productivity of an area for example by reducing the time and effort needed to
bring agricultural products to market, by providing access to schools, job opportunities and medical center. In the previous years there have been many debates whether transport infrastructures cause growth or economic growth cause transport infrastructures investment. But they all agree that there is high correlation between them. In this research it is assumed there is a positive relation between economic development and transport network performance.
Road network development has a major influence on the distribution of population, location of industries, exploitation of resources and provision of social service (Geurs K.T. & J.R, 2001). In long term it may stimulate land-use shifts, increase demand and behavioural shifts (M. Zuidgeest, 2005). Due to construction of new transport infrastructures new trips may be generated and people may change their mode of transportation. For example if a road is constructed in a place where the dominate mode of transportation is walking, due to the new road vehicles will start accessing that area. People may change their mode of transport from walking to public transport or privet cars.
Since road networks are designed for long term, understanding of how the existing network serves the different sectors of economy as well as identification of economically potential areas where lack of road network is holding back the development is essential. This can be done by analysing the accessibility of different opportunities, resources like mining, social services and schools.
